2023-09-24 21:46:46

by Philipp Hortmann

[permalink] [raw]
Subject: [PATCH 3/4] staging: rtl8192e: Unlock mutex for one line in rtllib_stop_protocol()

The following command uses the mutex ieee->wx_mutex:
cancel_delayed_work_sync(&ieee->associate_retry_wq)
Therefore it cannot be canceled with ieee->wx_mutex locked.
rtllib_stop_protocol() is always called with ieee->wx_mutex locked so
according line can be unlocked.

Signed-off-by: Philipp Hortmann <[email protected]>
---
[ 848.458088] ======================================================
[ 848.458089] WARNING: possible circular locking dependency detected
[ 848.458091] 6.6.0-rc1+ #15 Tainted: G C OE
[ 848.458102] ------------------------------------------------------
[ 848.458103] kworker/0:9/346 is trying to acquire lock:
[ 848.458104] ffff88817dc37aa0 (&ieee->wx_mutex){+.+.}-{4:4}, at: rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458121]
but task is already holding lock:
[ 848.458122] ffffc900017cfe38 ((work_completion)(&(&ieee->associate_retry_wq)->work)){+.+.}-{0:0}, at: process_scheduled_works+0x27f/0x580
[ 848.458129]
which lock already depends on the new lock.

[ 848.458130]
the existing dependency chain (in reverse order) is:
[ 848.458131]
-> #1 ((work_completion)(&(&ieee->associate_retry_wq)->work)){+.+.}-{0:0}:
[ 848.458134] __flush_work+0x6d/0x490
[ 848.458137] __cancel_work_timer+0x137/0x1c0
[ 848.458140] cancel_delayed_work_sync+0x13/0x20
[ 848.458142] rtllib_stop_protocol.part.0+0x49/0x120 [rtllib]
[ 848.458151] rtllib_stop_protocol+0x1c/0x30 [rtllib]
[ 848.458159] rtllib_wx_set_essid+0x12a/0x150 [rtllib]
[ 848.458167] _rtl92e_wx_set_essid+0x4e/0xa0 [r8192e_pci]
[ 848.458176] ioctl_standard_iw_point+0x2e6/0x390
[ 848.458180] ioctl_standard_call+0xaa/0xe0
[ 848.458183] wireless_process_ioctl+0x149/0x170
[ 848.458185] wext_handle_ioctl+0x9e/0x100
[ 848.458188] sock_ioctl+0x203/0x340
[ 848.458192] __x64_sys_ioctl+0x98/0xd0
[ 848.458195] do_syscall_64+0x3b/0x90
[ 848.458198] entry_SYSCALL_64_after_hwframe+0x6e/0xd8
[ 848.458202]
-> #0 (&ieee->wx_mutex){+.+.}-{4:4}:
[ 848.458206] __lock_acquire+0x12e0/0x1de0
[ 848.458209] lock_acquire+0xdc/0x2c0
[ 848.458212] __mutex_lock+0x99/0xce0
[ 848.458216] mutex_lock_nested+0x1b/0x30
[ 848.458219] rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458227] process_scheduled_works+0x308/0x580
[ 848.458229] worker_thread+0x19b/0x360
[ 848.458231] kthread+0x116/0x150
[ 848.458234] ret_from_fork+0x3c/0x60
[ 848.458237] ret_from_fork_asm+0x1b/0x30
[ 848.458242]
other info that might help us debug this:

[ 848.458243] Possible unsafe locking scenario:

[ 848.458244] CPU0 CPU1
[ 848.458245] ---- ----
[ 848.458246] lock((work_completion)(&(&ieee->associate_retry_wq)->work));
[ 848.458248] lock(&ieee->wx_mutex);
[ 848.458250] lock((work_completion)(&(&ieee->associate_retry_wq)->work));
[ 848.458252] lock(&ieee->wx_mutex);
[ 848.458254]
*** DEADLOCK ***

[ 848.458255] 2 locks held by kworker/0:9/346:
[ 848.458257] #0: ffff88810004f148 ((wq_completion)events){+.+.}-{0:0}, at: process_scheduled_works+0x27f/0x580
[ 848.458262] #1: ffffc900017cfe38 ((work_completion)(&(&ieee->associate_retry_wq)->work)){+.+.}-{0:0}, at: process_scheduled_works+0x27f/0x580
[ 848.458266]
stack backtrace:
[ 848.458268] CPU: 0 PID: 346 Comm: kworker/0:9 Tainted: G C OE 6.6.0-rc1+ #15
[ 848.458270] Hardware name: FUJITSU ESPRIMO P710/D3161-A1, BIOS V4.6.5.3 R1.16.0 for D3161-A1x 10/29/2012
[ 848.458272] Workqueue: events rtllib_associate_retry_wq [rtllib]
[ 848.458281] Call Trace:
[ 848.458283] <TASK>
[ 848.458285] dump_stack_lvl+0x5c/0xa0
[ 848.458288] dump_stack+0x10/0x20
[ 848.458291] print_circular_bug.isra.0+0x300/0x440
[ 848.458295] check_noncircular+0x136/0x150
[ 848.458298] ? ret_from_fork_asm+0x1b/0x30
[ 848.458302] ? __kernel_text_address+0x12/0x40
[ 848.458305] __lock_acquire+0x12e0/0x1de0
[ 848.458311] lock_acquire+0xdc/0x2c0
[ 848.458314] ? rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458324] __mutex_lock+0x99/0xce0
[ 848.458327] ? rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458336] ? rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458344] ? __this_cpu_preempt_check+0x13/0x20
[ 848.458348] mutex_lock_nested+0x1b/0x30
[ 848.458351] ? mutex_lock_nested+0x1b/0x30
[ 848.458354] rtllib_associate_retry_wq+0x2d/0xb0 [rtllib]
[ 848.458363] process_scheduled_works+0x308/0x580
[ 848.458368] ? __pfx_worker_thread+0x10/0x10
[ 848.458370] worker_thread+0x19b/0x360
[ 848.458373] ? __pfx_worker_thread+0x10/0x10
[ 848.458375] kthread+0x116/0x150
[ 848.458378] ? __pfx_kthread+0x10/0x10
[ 848.458381] ret_from_fork+0x3c/0x60
[ 848.458384] ? __pfx_kthread+0x10/0x10
[ 848.458387] ret_from_fork_asm+0x1b/0x30
[ 848.458392] </TASK>
---
drivers/staging/rtl8192e/rtllib_softmac.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/drivers/staging/rtl8192e/rtllib_softmac.c b/drivers/staging/rtl8192e/rtllib_softmac.c
index c874468d075c..72d0225dfdf1 100644
--- a/drivers/staging/rtl8192e/rtllib_softmac.c
+++ b/drivers/staging/rtl8192e/rtllib_softmac.c
@@ -2307,7 +2307,9 @@ void rtllib_stop_protocol(struct rtllib_device *ieee, u8 shutdown)
}

del_timer_sync(&ieee->associate_timer);
+ mutex_unlock(&ieee->wx_mutex);
cancel_delayed_work_sync(&ieee->associate_retry_wq);
+ mutex_lock(&ieee->wx_mutex);
cancel_delayed_work_sync(&ieee->link_change_wq);
rtllib_stop_scan(ieee);

--
2.42.0